Voltage measurement

[Rajinder1268]

Hi,
I have a requirement to measure voltages from a 10 pin header. I am interested in measuring 5V, 3V3 and 1V8 supplies with an accuracy of +/- 10mV. It is part of an automated test equipment, so really can't use a DVM.
If all the voltages are within range then indicate using a LED.
Not sure how to approach this.
Any help would be appreciated.

 

[KlausST]

Hi,

Where exactly is the problem? Ask a clear, elaborated question.

I mean some DVMs have a digital interface.

But you may also use ADC modules, maybe connected via USB.


Or you may use a microcontroller with built in or external ADC.
Basically what you need:
* ADC with accurate voltage reference
* maybe accurate enough voltage dividers
* I recommend a capacitor as LPF
* microcontroller with a little piece of software

Klaus

 

[stenzer]

Hi,

use a MCU and an ADC which supports a reference voltage / full-scale-range larger than 5 V, e.g. 6 V. Select an ADC with sufficient resolution, here a 12 Bit ADC in combination with a 6 V FSR should provide ~1.5 mV resolution, check the effective number of Bits of the ADC. Here, an ENOB of 10 Bit should be sufficient BUT this is only your resolution.

To give a feedback on the accuracy we need more information about the voltages which should be masured e.g. are this supply voltages or sensor signals. Temperature might be also of interest. Depending on this information a signal conditioning might be required e.g. a buffering structure which effects the accuracy. E.g. introducing an offset, which might be compensated by an offset (gain) calibration.

BR

 

[Rajinder1268]

Hi,
I was thinking of a microcontroller/ ADC option. The 3V3 is used to power an LCD and also other peripherals. The 5V is for CAN Comms, 1V8 is microcontroller core.

What would be the input circuit conditioning for the ADC? Just an LPF? Also how can I determine the values of RC.

 

[danadakk]

Do you code ? Eg. C code for example ?


Regards, Dana.

 

[Rajinder1268]

Do you code ? Eg. C code for example ?


Regards, Dana.

Hi, yes I can code in C.

 

[stenzer]

Without any further knowledge, a resitor divider will do the job, as mentioned by @KlausST

For the LPF the ADC sampling frequency has to be considered (fc < fs/2), as well as the frequencies expected to be observed or supressed within your observed voltages.

 

[danadakk]

Using simple block language, mBlock, and an ATTINY85 8 pin dip micro :

You need R divider on 5V input. You can ad averaging to get noise suppression on rail.

To program the ATTINY85 you do it with a Arduino UNO or Nano board, used as in circuit
programmer.

mBlock converts your block code to Arduino, and you use the Arduino IDE
to program the part.


Regards, Dana.

--- Updated Apr 5, 2023 ---

Alternatively use a PSOC, which is a single chip SOC, routable :

Shown above basic A/D or use one of the 3 power supervision components (a component in
PSOC is an internal resource).

Also there are DACs so you can do trim if needed on external supplies.

Here is what is on a PSOC, multiple copies of these components in many cases :

Each component has a rich set of libraries, basically f() calls, to manipulate the component.

Note onboard Vref is good to +/- .1%

IDE, PSOC Creator, and Compiler free. Good board for this would be $15, CY8CKIT-059.


Regards, Dana.

 

[KlausST]

Hi,

if it´s just to validate power supplies, why not using a dedicated IC?
Didn´t you do a part search before? Power supply monitors are quite common.

Many manufacturers. One of them: https://www.analog.com/en/product-category/supervisory-circuits.html

Klaus

 

[danadakk]

Note in post #8 you could also pulse led according to which supply
is posting a problem, just a thought. Or put out a message on a low
cost display. In PSOC case even do a PC alert over USBUART......

Either solution could do led approach.


Regards, Dana.

 

[crutschow]

how can I determine the values of RC.

It largely depends upon how fast you need to take each reading.
Typically you want as large an RC value as possible to suppress any noise but, to settle within 1mV of 5V (worst-case voltage), requires about 7 time-constants, so that would be the maximum.

 

[Rajinder1268]

It largely depends upon how fast you need to take each reading.
Typically you want as large an RC value as possible to suppress any noise but, to settle within 1mV of 5V (worst-case voltage), requires about 7 time-constants, so that would be the maximum.

Would I need to have a unity gain buffer before my RC filter?

 

[KlausST]

Hi,

Would I need to have a unity gain buffer before my RC filter?

I can't think of a benefit in your case...
It just introduces errors.

Klaus

 

[crutschow]

Would I need to have a unity gain buffer before my RC filter?

Not unless the source is insufficient to drive the RC filter, which is unlikely.